AArch64/ARM64: Port NEON post-increment load/store with 2/3/4 vectors to ARM64 backend.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@208284 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/ARM64/ARM64ISelDAGToDAG.cpp b/lib/Target/ARM64/ARM64ISelDAGToDAG.cpp
index d9c945ccd920fb5fb3545bd727d2f09b6abad173..f216f79255e44c55b9dac39eb2d82b68f7489a89 100644 (file)
--- a/lib/Target/ARM64/ARM64ISelDAGToDAG.cpp
+++ b/lib/Target/ARM64/ARM64ISelDAGToDAG.cpp
@@ -150,10 +150,15 @@ public:
 
   SDNode *SelectLoad(SDNode *N, unsigned NumVecs, unsigned Opc,
                      unsigned SubRegIdx);
+  SDNode *SelectPostLoad(SDNode *N, unsigned NumVecs, unsigned Opc,
+                         unsigned SubRegIdx);
   SDNode *SelectLoadLane(SDNode *N, unsigned NumVecs, unsigned Opc);
+  SDNode *SelectPostLoadLane(SDNode *N, unsigned NumVecs, unsigned Opc);
 
   SDNode *SelectStore(SDNode *N, unsigned NumVecs, unsigned Opc);
+  SDNode *SelectPostStore(SDNode *N, unsigned NumVecs, unsigned Opc);
   SDNode *SelectStoreLane(SDNode *N, unsigned NumVecs, unsigned Opc);
+  SDNode *SelectPostStoreLane(SDNode *N, unsigned NumVecs, unsigned Opc);
 
   SDNode *SelectSIMDAddSubNarrowing(unsigned IntNo, SDNode *Node);
   SDNode *SelectSIMDXtnNarrowing(unsigned IntNo, SDNode *Node);
@@ -952,33 +957,43 @@ SDNode *ARM64DAGToDAGISel::SelectLoad(SDNode *N, unsigned NumVecs, unsigned Opc,
 
   SDNode *Ld = CurDAG->getMachineNode(Opc, dl, ResTys, Ops);
   SDValue SuperReg = SDValue(Ld, 0);
-
-  // MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
-  // MemOp[0] = cast<MemIntrinsicSDNode>(N)->getMemOperand();
-  // cast<MachineSDNode>(Ld)->setMemRefs(MemOp, MemOp + 1);
-
-  switch (NumVecs) {
-  case 4:
-    ReplaceUses(SDValue(N, 3), CurDAG->getTargetExtractSubreg(SubRegIdx + 3, dl,
-                                                              VT, SuperReg));
-  // FALLTHROUGH
-  case 3:
-    ReplaceUses(SDValue(N, 2), CurDAG->getTargetExtractSubreg(SubRegIdx + 2, dl,
-                                                              VT, SuperReg));
-  // FALLTHROUGH
-  case 2:
-    ReplaceUses(SDValue(N, 1), CurDAG->getTargetExtractSubreg(SubRegIdx + 1, dl,
-                                                              VT, SuperReg));
-    ReplaceUses(SDValue(N, 0),
-                CurDAG->getTargetExtractSubreg(SubRegIdx, dl, VT, SuperReg));
-    break;
-  case 1:
-    ReplaceUses(SDValue(N, 0), SuperReg);
-    break;
-  }
+  for (unsigned i = 0; i < NumVecs; ++i)
+    ReplaceUses(SDValue(N, i),
+        CurDAG->getTargetExtractSubreg(SubRegIdx + i, dl, VT, SuperReg));
 
   ReplaceUses(SDValue(N, NumVecs), SDValue(Ld, 1));
+  return nullptr;
+}
+
+SDNode *ARM64DAGToDAGISel::SelectPostLoad(SDNode *N, unsigned NumVecs,
+                                          unsigned Opc, unsigned SubRegIdx) {
+  SDLoc dl(N);
+  EVT VT = N->getValueType(0);
+  SDValue Chain = N->getOperand(0);
+
+  SmallVector<SDValue, 6> Ops;
+  Ops.push_back(N->getOperand(1)); // Mem operand
+  Ops.push_back(N->getOperand(2)); // Incremental
+  Ops.push_back(Chain);
+
+  std::vector<EVT> ResTys;
+  ResTys.push_back(MVT::i64); // Type of the write back register
+  ResTys.push_back(MVT::Untyped);
+  ResTys.push_back(MVT::Other);
+
+  SDNode *Ld = CurDAG->getMachineNode(Opc, dl, ResTys, Ops);
+
+  // Update uses of write back register
+  ReplaceUses(SDValue(N, NumVecs), SDValue(Ld, 0));
+
+  // Update uses of vector list
+  SDValue SuperReg = SDValue(Ld, 1);
+  for (unsigned i = 0; i < NumVecs; ++i)
+    ReplaceUses(SDValue(N, i),
+        CurDAG->getTargetExtractSubreg(SubRegIdx + i, dl, VT, SuperReg));
 
+  // Update the chain
+  ReplaceUses(SDValue(N, NumVecs + 1), SDValue(Ld, 2));
   return nullptr;
 }
 
@@ -1001,6 +1016,29 @@ SDNode *ARM64DAGToDAGISel::SelectStore(SDNode *N, unsigned NumVecs,
   return St;
 }
 
+SDNode *ARM64DAGToDAGISel::SelectPostStore(SDNode *N, unsigned NumVecs,
+                                               unsigned Opc) {
+  SDLoc dl(N);
+  EVT VT = N->getOperand(2)->getValueType(0);
+  SmallVector<EVT, 2> ResTys;
+  ResTys.push_back(MVT::i64);   // Type of the write back register
+  ResTys.push_back(MVT::Other); // Type for the Chain
+
+  // Form a REG_SEQUENCE to force register allocation.
+  bool Is128Bit = VT.getSizeInBits() == 128;
+  SmallVector<SDValue, 4> Regs(N->op_begin() + 1, N->op_begin() + 1 + NumVecs);
+  SDValue RegSeq = Is128Bit ? createQTuple(Regs) : createDTuple(Regs);
+
+  SmallVector<SDValue, 6> Ops;
+  Ops.push_back(RegSeq);
+  Ops.push_back(N->getOperand(NumVecs + 1)); // base register
+  Ops.push_back(N->getOperand(NumVecs + 2)); // Incremental
+  Ops.push_back(N->getOperand(0)); // Chain
+  SDNode *St = CurDAG->getMachineNode(Opc, dl, ResTys, Ops);
+
+  return St;
+}
+
 /// WidenVector - Given a value in the V64 register class, produce the
 /// equivalent value in the V128 register class.
 class WidenVector {
@@ -1065,42 +1103,68 @@ SDNode *ARM64DAGToDAGISel::SelectLoadLane(SDNode *N, unsigned NumVecs,
   SDValue SuperReg = SDValue(Ld, 0);
 
   EVT WideVT = RegSeq.getOperand(1)->getValueType(0);
-  switch (NumVecs) {
-  case 4: {
-    SDValue NV3 =
-        CurDAG->getTargetExtractSubreg(ARM64::qsub3, dl, WideVT, SuperReg);
+  static unsigned QSubs[] = { ARM64::qsub0, ARM64::qsub1, ARM64::qsub2,
+                              ARM64::qsub3 };
+  for (unsigned i = 0; i < NumVecs; ++i) {
+    SDValue NV = CurDAG->getTargetExtractSubreg(QSubs[i], dl, WideVT, SuperReg);
     if (Narrow)
-      ReplaceUses(SDValue(N, 3), NarrowVector(NV3, *CurDAG));
-    else
-      ReplaceUses(SDValue(N, 3), NV3);
+      NV = NarrowVector(NV, *CurDAG);
+    ReplaceUses(SDValue(N, i), NV);
   }
-  // FALLTHROUGH
-  case 3: {
-    SDValue NV2 =
-        CurDAG->getTargetExtractSubreg(ARM64::qsub2, dl, WideVT, SuperReg);
+
+  ReplaceUses(SDValue(N, NumVecs), SDValue(Ld, 1));
+
+  return Ld;
+}
+
+SDNode *ARM64DAGToDAGISel::SelectPostLoadLane(SDNode *N, unsigned NumVecs,
+                                              unsigned Opc) {
+  SDLoc dl(N);
+  EVT VT = N->getValueType(0);
+  bool Narrow = VT.getSizeInBits() == 64;
+
+  // Form a REG_SEQUENCE to force register allocation.
+  SmallVector<SDValue, 4> Regs(N->op_begin() + 1, N->op_begin() + 1 + NumVecs);
+
+  if (Narrow)
+    std::transform(Regs.begin(), Regs.end(), Regs.begin(),
+                   WidenVector(*CurDAG));
+
+  SDValue RegSeq = createQTuple(Regs);
+
+  std::vector<EVT> ResTys;
+  ResTys.push_back(MVT::i64); // Type of the write back register
+  ResTys.push_back(MVT::Untyped);
+  ResTys.push_back(MVT::Other);
+
+  unsigned LaneNo =
+      cast<ConstantSDNode>(N->getOperand(NumVecs + 1))->getZExtValue();
+
+  SmallVector<SDValue, 6> Ops;
+  Ops.push_back(RegSeq);
+  Ops.push_back(CurDAG->getTargetConstant(LaneNo, MVT::i64)); // Lane Number
+  Ops.push_back(N->getOperand(NumVecs + 2)); // Base register
+  Ops.push_back(N->getOperand(NumVecs + 3)); // Incremental
+  Ops.push_back(N->getOperand(0));
+  SDNode *Ld = CurDAG->getMachineNode(Opc, dl, ResTys, Ops);
+
+  // Update uses of the write back register
+  ReplaceUses(SDValue(N, NumVecs), SDValue(Ld, 0));
+
+  // Update uses of the vector list
+  SDValue SuperReg = SDValue(Ld, 1);
+  EVT WideVT = RegSeq.getOperand(1)->getValueType(0);
+  static unsigned QSubs[] = { ARM64::qsub0, ARM64::qsub1, ARM64::qsub2,
+                              ARM64::qsub3 };
+  for (unsigned i = 0; i < NumVecs; ++i) {
+    SDValue NV = CurDAG->getTargetExtractSubreg(QSubs[i], dl, WideVT, SuperReg);
     if (Narrow)
-      ReplaceUses(SDValue(N, 2), NarrowVector(NV2, *CurDAG));
-    else
-      ReplaceUses(SDValue(N, 2), NV2);
-  }
-  // FALLTHROUGH
-  case 2: {
-    SDValue NV1 =
-        CurDAG->getTargetExtractSubreg(ARM64::qsub1, dl, WideVT, SuperReg);
-    SDValue NV0 =
-        CurDAG->getTargetExtractSubreg(ARM64::qsub0, dl, WideVT, SuperReg);
-    if (Narrow) {
-      ReplaceUses(SDValue(N, 1), NarrowVector(NV1, *CurDAG));
-      ReplaceUses(SDValue(N, 0), NarrowVector(NV0, *CurDAG));
-    } else {
-      ReplaceUses(SDValue(N, 1), NV1);
-      ReplaceUses(SDValue(N, 0), NV0);
-    }
-    break;
-  }
+      NV = NarrowVector(NV, *CurDAG);
+    ReplaceUses(SDValue(N, i), NV);
   }
 
-  ReplaceUses(SDValue(N, NumVecs), SDValue(Ld, 1));
+  // Update the Chain
+  ReplaceUses(SDValue(N, NumVecs + 1), SDValue(Ld, 2));
 
   return Ld;
 }
@@ -1138,6 +1202,44 @@ SDNode *ARM64DAGToDAGISel::SelectStoreLane(SDNode *N, unsigned NumVecs,
   return St;
 }
 
+SDNode *ARM64DAGToDAGISel::SelectPostStoreLane(SDNode *N, unsigned NumVecs,
+                                               unsigned Opc) {
+  SDLoc dl(N);
+  EVT VT = N->getOperand(2)->getValueType(0);
+  bool Narrow = VT.getSizeInBits() == 64;
+
+  // Form a REG_SEQUENCE to force register allocation.
+  SmallVector<SDValue, 4> Regs(N->op_begin() + 1, N->op_begin() + 1 + NumVecs);
+
+  if (Narrow)
+    std::transform(Regs.begin(), Regs.end(), Regs.begin(),
+                   WidenVector(*CurDAG));
+
+  SDValue RegSeq = createQTuple(Regs);
+
+  SmallVector<EVT, 2> ResTys;
+  ResTys.push_back(MVT::i64);   // Type of the write back register
+  ResTys.push_back(MVT::Other);
+
+  unsigned LaneNo =
+      cast<ConstantSDNode>(N->getOperand(NumVecs + 1))->getZExtValue();
+
+  SmallVector<SDValue, 6> Ops;
+  Ops.push_back(RegSeq);
+  Ops.push_back(CurDAG->getTargetConstant(LaneNo, MVT::i64));
+  Ops.push_back(N->getOperand(NumVecs + 2)); // Base Register
+  Ops.push_back(N->getOperand(NumVecs + 3)); // Incremental
+  Ops.push_back(N->getOperand(0));
+  SDNode *St = CurDAG->getMachineNode(Opc, dl, ResTys, Ops);
+
+  // Transfer memoperands.
+  MachineSDNode::mmo_iterator MemOp = MF->allocateMemRefsArray(1);
+  MemOp[0] = cast<MemIntrinsicSDNode>(N)->getMemOperand();
+  cast<MachineSDNode>(St)->setMemRefs(MemOp, MemOp + 1);
+
+  return St;
+}
+
 static bool isBitfieldExtractOpFromAnd(SelectionDAG *CurDAG, SDNode *N,
                                        unsigned &Opc, SDValue &Opd0,
                                        unsigned &LSB, unsigned &MSB,
@@ -2441,6 +2543,378 @@ SDNode *ARM64DAGToDAGISel::Select(SDNode *Node) {
     }
     }
   }
+  case ARM64ISD::LD2post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 2, ARM64::LD2Twov2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD3post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 3, ARM64::LD3Threev2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD4post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 4, ARM64::LD4Fourv2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD1x2post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 2, ARM64::LD1Twov2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD1x3post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 3, ARM64::LD1Threev2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD1x4post: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 4, ARM64::LD1Fourv2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD2DUPpost: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 2, ARM64::LD2Rv2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD3DUPpost: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 3, ARM64::LD3Rv2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD4DUPpost: {
+    if (VT == MVT::v8i8)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv8b_POST, ARM64::dsub0);
+    else if (VT == MVT::v16i8)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv16b_POST, ARM64::qsub0);
+    else if (VT == MVT::v4i16)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv4h_POST, ARM64::dsub0);
+    else if (VT == MVT::v8i16)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv8h_POST, ARM64::qsub0);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv2s_POST, ARM64::dsub0);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv4s_POST, ARM64::qsub0);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv1d_POST, ARM64::dsub0);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostLoad(Node, 4, ARM64::LD4Rv2d_POST, ARM64::qsub0);
+    break;
+  }
+  case ARM64ISD::LD2LANEpost: {
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostLoadLane(Node, 2, ARM64::LD2i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostLoadLane(Node, 2, ARM64::LD2i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostLoadLane(Node, 2, ARM64::LD2i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostLoadLane(Node, 2, ARM64::LD2i64_POST);
+    break;
+  }
+  case ARM64ISD::LD3LANEpost: {
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostLoadLane(Node, 3, ARM64::LD3i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostLoadLane(Node, 3, ARM64::LD3i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostLoadLane(Node, 3, ARM64::LD3i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostLoadLane(Node, 3, ARM64::LD3i64_POST);
+    break;
+  }
+  case ARM64ISD::LD4LANEpost: {
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostLoadLane(Node, 4, ARM64::LD4i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostLoadLane(Node, 4, ARM64::LD4i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostLoadLane(Node, 4, ARM64::LD4i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostLoadLane(Node, 4, ARM64::LD4i64_POST);
+    break;
+  }
+  case ARM64ISD::ST2post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov4s_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 2, ARM64::ST2Twov2d_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov1d_POST);
+    break;
+  }
+  case ARM64ISD::ST3post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev4s_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 3, ARM64::ST3Threev2d_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev1d_POST);
+    break;
+  }
+  case ARM64ISD::ST4post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv4s_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 4, ARM64::ST4Fourv2d_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv1d_POST);
+    break;
+  }
+  case ARM64ISD::ST1x2post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov4s_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov1d_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 2, ARM64::ST1Twov2d_POST);
+    break;
+  }
+  case ARM64ISD::ST1x3post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev4s_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev1d_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 3, ARM64::ST1Threev2d_POST);
+    break;
+  }
+  case ARM64ISD::ST1x4post: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v8i8)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv8b_POST);
+    else if (VT == MVT::v16i8)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv16b_POST);
+    else if (VT == MVT::v4i16)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv4h_POST);
+    else if (VT == MVT::v8i16)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv8h_POST);
+    else if (VT == MVT::v2i32 || VT == MVT::v2f32)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv2s_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v4f32)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv4s_POST);
+    else if (VT == MVT::v1i64 || VT == MVT::v1f64)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv1d_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v2f64)
+      return SelectPostStore(Node, 4, ARM64::ST1Fourv2d_POST);
+    break;
+  }
+  case ARM64ISD::ST2LANEpost: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostStoreLane(Node, 2, ARM64::ST2i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostStoreLane(Node, 2, ARM64::ST2i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostStoreLane(Node, 2, ARM64::ST2i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostStoreLane(Node, 2, ARM64::ST2i64_POST);
+    break;
+  }
+  case ARM64ISD::ST3LANEpost: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostStoreLane(Node, 3, ARM64::ST3i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostStoreLane(Node, 3, ARM64::ST3i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostStoreLane(Node, 3, ARM64::ST3i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostStoreLane(Node, 3, ARM64::ST3i64_POST);
+    break;
+  }
+  case ARM64ISD::ST4LANEpost: {
+    VT = Node->getOperand(1).getValueType();
+    if (VT == MVT::v16i8 || VT == MVT::v8i8)
+      return SelectPostStoreLane(Node, 4, ARM64::ST4i8_POST);
+    else if (VT == MVT::v8i16 || VT == MVT::v4i16)
+      return SelectPostStoreLane(Node, 4, ARM64::ST4i16_POST);
+    else if (VT == MVT::v4i32 || VT == MVT::v2i32 || VT == MVT::v4f32 ||
+             VT == MVT::v2f32)
+      return SelectPostStoreLane(Node, 4, ARM64::ST4i32_POST);
+    else if (VT == MVT::v2i64 || VT == MVT::v1i64 || VT == MVT::v2f64 ||
+             VT == MVT::v1f64)
+      return SelectPostStoreLane(Node, 4, ARM64::ST4i64_POST);
+    break;
+  }
 
   case ISD::FCEIL:
   case ISD::FFLOOR:

diff --git a/lib/Target/ARM64/ARM64ISelLowering.cpp b/lib/Target/ARM64/ARM64ISelLowering.cpp
index 4c6d7648d578521d8ade9c0ce34c9a696be54977..bff6ba060fb5de08beede8cb34a19078abc9ba60 100644 (file)
--- a/lib/Target/ARM64/ARM64ISelLowering.cpp
+++ b/lib/Target/ARM64/ARM64ISelLowering.cpp
@@ -369,6 +369,9 @@ ARM64TargetLowering::ARM64TargetLowering(ARM64TargetMachine &TM)
   setTargetDAGCombine(ISD::SELECT);
   setTargetDAGCombine(ISD::VSELECT);
 
+  setTargetDAGCombine(ISD::INTRINSIC_VOID);
+  setTargetDAGCombine(ISD::INTRINSIC_W_CHAIN);
+
   MaxStoresPerMemset = MaxStoresPerMemsetOptSize = 8;
   MaxStoresPerMemcpy = MaxStoresPerMemcpyOptSize = 4;
   MaxStoresPerMemmove = MaxStoresPerMemmoveOptSize = 4;
@@ -729,6 +732,27 @@ const char *ARM64TargetLowering::getTargetNodeName(unsigned Opcode) const {
   case ARM64ISD::URSHR_I:           return "ARM64ISD::URSHR_I";
   case ARM64ISD::SQSHLU_I:          return "ARM64ISD::SQSHLU_I";
   case ARM64ISD::WrapperLarge:      return "ARM64ISD::WrapperLarge";
+  case ARM64ISD::LD2post:           return "ARM64ISD::LD2post";
+  case ARM64ISD::LD3post:           return "ARM64ISD::LD3post";
+  case ARM64ISD::LD4post:           return "ARM64ISD::LD4post";
+  case ARM64ISD::ST2post:           return "ARM64ISD::ST2post";
+  case ARM64ISD::ST3post:           return "ARM64ISD::ST3post";
+  case ARM64ISD::ST4post:           return "ARM64ISD::ST4post";
+  case ARM64ISD::LD1x2post:         return "ARM64ISD::LD1x2post";
+  case ARM64ISD::LD1x3post:         return "ARM64ISD::LD1x3post";
+  case ARM64ISD::LD1x4post:         return "ARM64ISD::LD1x4post";
+  case ARM64ISD::ST1x2post:         return "ARM64ISD::ST1x2post";
+  case ARM64ISD::ST1x3post:         return "ARM64ISD::ST1x3post";
+  case ARM64ISD::ST1x4post:         return "ARM64ISD::ST1x4post";
+  case ARM64ISD::LD2DUPpost:        return "ARM64ISD::LD2DUPpost";
+  case ARM64ISD::LD3DUPpost:        return "ARM64ISD::LD3DUPpost";
+  case ARM64ISD::LD4DUPpost:        return "ARM64ISD::LD4DUPpost";
+  case ARM64ISD::LD2LANEpost:       return "ARM64ISD::LD2LANEpost";
+  case ARM64ISD::LD3LANEpost:       return "ARM64ISD::LD3LANEpost";
+  case ARM64ISD::LD4LANEpost:       return "ARM64ISD::LD4LANEpost";
+  case ARM64ISD::ST2LANEpost:       return "ARM64ISD::ST2LANEpost";
+  case ARM64ISD::ST3LANEpost:       return "ARM64ISD::ST3LANEpost";
+  case ARM64ISD::ST4LANEpost:       return "ARM64ISD::ST4LANEpost";
   }
 }
 
@@ -5683,6 +5707,9 @@ bool ARM64TargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
   case Intrinsic::arm64_neon_ld2:
   case Intrinsic::arm64_neon_ld3:
   case Intrinsic::arm64_neon_ld4:
+  case Intrinsic::arm64_neon_ld1x2:
+  case Intrinsic::arm64_neon_ld1x3:
+  case Intrinsic::arm64_neon_ld1x4:
   case Intrinsic::arm64_neon_ld2lane:
   case Intrinsic::arm64_neon_ld3lane:
   case Intrinsic::arm64_neon_ld4lane:
@@ -5704,6 +5731,9 @@ bool ARM64TargetLowering::getTgtMemIntrinsic(IntrinsicInfo &Info,
   case Intrinsic::arm64_neon_st2:
   case Intrinsic::arm64_neon_st3:
   case Intrinsic::arm64_neon_st4:
+  case Intrinsic::arm64_neon_st1x2:
+  case Intrinsic::arm64_neon_st1x3:
+  case Intrinsic::arm64_neon_st1x4:
   case Intrinsic::arm64_neon_st2lane:
   case Intrinsic::arm64_neon_st3lane:
   case Intrinsic::arm64_neon_st4lane: {
@@ -7038,6 +7068,138 @@ static SDValue performSTORECombine(SDNode *N,
                       S->getAlignment());
 }
 
+/// Target-specific DAG combine function for NEON load/store intrinsics
+/// to merge base address updates.
+static SDValue performNEONPostLDSTCombine(SDNode *N,
+                                          TargetLowering::DAGCombinerInfo &DCI,
+                                          SelectionDAG &DAG) {
+  if (DCI.isBeforeLegalize() || DCI.isCalledByLegalizer())
+    return SDValue();
+
+  unsigned AddrOpIdx = N->getNumOperands() - 1;
+  SDValue Addr = N->getOperand(AddrOpIdx);
+
+  // Search for a use of the address operand that is an increment.
+  for (SDNode::use_iterator UI = Addr.getNode()->use_begin(),
+       UE = Addr.getNode()->use_end(); UI != UE; ++UI) {
+    SDNode *User = *UI;
+    if (User->getOpcode() != ISD::ADD ||
+        UI.getUse().getResNo() != Addr.getResNo())
+      continue;
+
+    // Check that the add is independent of the load/store.  Otherwise, folding
+    // it would create a cycle.
+    if (User->isPredecessorOf(N) || N->isPredecessorOf(User))
+      continue;
+
+    // Find the new opcode for the updating load/store.
+    bool IsStore = false;
+    bool IsLaneOp = false;
+    bool IsDupOp = false;
+    unsigned NewOpc = 0;
+    unsigned NumVecs = 0;
+    unsigned IntNo = cast<ConstantSDNode>(N->getOperand(1))->getZExtValue();
+    switch (IntNo) {
+    default: llvm_unreachable("unexpected intrinsic for Neon base update");
+    case Intrinsic::arm64_neon_ld2:       NewOpc = ARM64ISD::LD2post;
+      NumVecs = 2; break;
+    case Intrinsic::arm64_neon_ld3:       NewOpc = ARM64ISD::LD3post;
+      NumVecs = 3; break;
+    case Intrinsic::arm64_neon_ld4:       NewOpc = ARM64ISD::LD4post;
+      NumVecs = 4; break;
+    case Intrinsic::arm64_neon_st2:       NewOpc = ARM64ISD::ST2post;
+      NumVecs = 2; IsStore = true; break;
+    case Intrinsic::arm64_neon_st3:       NewOpc = ARM64ISD::ST3post;
+      NumVecs = 3; IsStore = true; break;
+    case Intrinsic::arm64_neon_st4:       NewOpc = ARM64ISD::ST4post;
+      NumVecs = 4; IsStore = true; break;
+    case Intrinsic::arm64_neon_ld1x2:     NewOpc = ARM64ISD::LD1x2post;
+      NumVecs = 2; break;
+    case Intrinsic::arm64_neon_ld1x3:     NewOpc = ARM64ISD::LD1x3post;
+      NumVecs = 3; break;
+    case Intrinsic::arm64_neon_ld1x4:     NewOpc = ARM64ISD::LD1x4post;
+      NumVecs = 4; break;
+    case Intrinsic::arm64_neon_st1x2:     NewOpc = ARM64ISD::ST1x2post;
+      NumVecs = 2; IsStore = true; break;
+    case Intrinsic::arm64_neon_st1x3:     NewOpc = ARM64ISD::ST1x3post;
+      NumVecs = 3; IsStore = true; break;
+    case Intrinsic::arm64_neon_st1x4:     NewOpc = ARM64ISD::ST1x4post;
+      NumVecs = 4; IsStore = true; break;
+    case Intrinsic::arm64_neon_ld2r:      NewOpc = ARM64ISD::LD2DUPpost;
+      NumVecs = 2; IsDupOp = true; break;
+    case Intrinsic::arm64_neon_ld3r:      NewOpc = ARM64ISD::LD3DUPpost;
+      NumVecs = 3; IsDupOp = true; break;
+    case Intrinsic::arm64_neon_ld4r:      NewOpc = ARM64ISD::LD4DUPpost;
+      NumVecs = 4; IsDupOp = true; break;
+    case Intrinsic::arm64_neon_ld2lane:   NewOpc = ARM64ISD::LD2LANEpost;
+      NumVecs = 2; IsLaneOp = true; break;
+    case Intrinsic::arm64_neon_ld3lane:   NewOpc = ARM64ISD::LD3LANEpost;
+      NumVecs = 3; IsLaneOp = true; break;
+    case Intrinsic::arm64_neon_ld4lane:   NewOpc = ARM64ISD::LD4LANEpost;
+      NumVecs = 4; IsLaneOp = true; break;
+    case Intrinsic::arm64_neon_st2lane:   NewOpc = ARM64ISD::ST2LANEpost;
+      NumVecs = 2; IsStore = true; IsLaneOp = true; break;
+    case Intrinsic::arm64_neon_st3lane:   NewOpc = ARM64ISD::ST3LANEpost;
+      NumVecs = 3; IsStore = true; IsLaneOp = true; break;
+    case Intrinsic::arm64_neon_st4lane:   NewOpc = ARM64ISD::ST4LANEpost;
+      NumVecs = 4; IsStore = true; IsLaneOp = true; break;
+    }
+
+    EVT VecTy;
+    if (IsStore)
+      VecTy = N->getOperand(2).getValueType();
+    else
+      VecTy = N->getValueType(0);
+
+    // If the increment is a constant, it must match the memory ref size.
+    SDValue Inc = User->getOperand(User->getOperand(0) == Addr ? 1 : 0);
+    if (ConstantSDNode *CInc = dyn_cast<ConstantSDNode>(Inc.getNode())) {
+      uint32_t IncVal = CInc->getZExtValue();
+      unsigned NumBytes = NumVecs * VecTy.getSizeInBits() / 8;
+      if (IsLaneOp || IsDupOp)
+        NumBytes /= VecTy.getVectorNumElements();
+      if (IncVal != NumBytes)
+        continue;
+      Inc = DAG.getRegister(ARM64::XZR, MVT::i64);
+    }
+    SmallVector<SDValue, 8> Ops;
+    Ops.push_back(N->getOperand(0)); // Incoming chain
+    // Load lane and store have vector list as input.
+    if (IsLaneOp || IsStore)
+      for (unsigned i = 2; i < AddrOpIdx; ++i)
+        Ops.push_back(N->getOperand(i));
+    Ops.push_back(N->getOperand(AddrOpIdx)); // Base register
+    Ops.push_back(Inc);
+
+    // Return Types.
+    EVT Tys[6];
+    unsigned NumResultVecs = (IsStore ? 0 : NumVecs);
+    unsigned n;
+    for (n = 0; n < NumResultVecs; ++n)
+      Tys[n] = VecTy;
+    Tys[n++] = MVT::i64;  // Type of write back register
+    Tys[n] = MVT::Other;  // Type of the chain
+    SDVTList SDTys = DAG.getVTList(ArrayRef<EVT>(Tys, NumResultVecs + 2));
+
+    MemIntrinsicSDNode *MemInt = cast<MemIntrinsicSDNode>(N);
+    SDValue UpdN = DAG.getMemIntrinsicNode(NewOpc, SDLoc(N), SDTys, Ops,
+                                           MemInt->getMemoryVT(),
+                                           MemInt->getMemOperand());
+
+    // Update the uses.
+    std::vector<SDValue> NewResults;
+    for (unsigned i = 0; i < NumResultVecs; ++i) {
+      NewResults.push_back(SDValue(UpdN.getNode(), i));
+    }
+    NewResults.push_back(SDValue(UpdN.getNode(), NumResultVecs + 1));
+    DCI.CombineTo(N, NewResults);
+    DCI.CombineTo(User, SDValue(UpdN.getNode(), NumResultVecs));
+
+    break;
+  }
+  return SDValue();
+}
+
 // Optimize compare with zero and branch.
 static SDValue performBRCONDCombine(SDNode *N,
                                     TargetLowering::DAGCombinerInfo &DCI,
@@ -7196,6 +7358,34 @@ SDValue ARM64TargetLowering::PerformDAGCombine(SDNode *N,
     return performSTORECombine(N, DCI, DAG, Subtarget);
   case ARM64ISD::BRCOND:
     return performBRCONDCombine(N, DCI, DAG);
+  case ISD::INTRINSIC_VOID:
+  case ISD::INTRINSIC_W_CHAIN:
+    switch (cast<ConstantSDNode>(N->getOperand(1))->getZExtValue()) {
+    case Intrinsic::arm64_neon_ld2:
+    case Intrinsic::arm64_neon_ld3:
+    case Intrinsic::arm64_neon_ld4:
+    case Intrinsic::arm64_neon_ld1x2:
+    case Intrinsic::arm64_neon_ld1x3:
+    case Intrinsic::arm64_neon_ld1x4:
+    case Intrinsic::arm64_neon_ld2lane:
+    case Intrinsic::arm64_neon_ld3lane:
+    case Intrinsic::arm64_neon_ld4lane:
+    case Intrinsic::arm64_neon_ld2r:
+    case Intrinsic::arm64_neon_ld3r:
+    case Intrinsic::arm64_neon_ld4r:
+    case Intrinsic::arm64_neon_st2:
+    case Intrinsic::arm64_neon_st3:
+    case Intrinsic::arm64_neon_st4:
+    case Intrinsic::arm64_neon_st1x2:
+    case Intrinsic::arm64_neon_st1x3:
+    case Intrinsic::arm64_neon_st1x4:
+    case Intrinsic::arm64_neon_st2lane:
+    case Intrinsic::arm64_neon_st3lane:
+    case Intrinsic::arm64_neon_st4lane:
+      return performNEONPostLDSTCombine(N, DCI, DAG);
+    default:
+      break;
+    }
   }
   return SDValue();
 }

diff --git a/lib/Target/ARM64/ARM64ISelLowering.h b/lib/Target/ARM64/ARM64ISelLowering.h
index aa27b2d43d24922968c3cb508dfafeba1aef0dfe..8b321ee9d03d17351b93d35cae2d08feb32dd338 100644 (file)
--- a/lib/Target/ARM64/ARM64ISelLowering.h
+++ b/lib/Target/ARM64/ARM64ISelLowering.h
@@ -160,7 +160,30 @@ enum {
 
   // {s|u}int to FP within a FP register.
   SITOF,
-  UITOF
+  UITOF,
+
+  // NEON Load/Store with post-increment base updates
+  LD2post = ISD::FIRST_TARGET_MEMORY_OPCODE,
+  LD3post,
+  LD4post,
+  ST2post,
+  ST3post,
+  ST4post,
+  LD1x2post,
+  LD1x3post,
+  LD1x4post,
+  ST1x2post,
+  ST1x3post,
+  ST1x4post,
+  LD2DUPpost,
+  LD3DUPpost,
+  LD4DUPpost,
+  LD2LANEpost,
+  LD3LANEpost,
+  LD4LANEpost,
+  ST2LANEpost,
+  ST3LANEpost,
+  ST4LANEpost
 };
 
 } // end namespace ARM64ISD

diff --git a/test/CodeGen/ARM64/indexed-vector-ldst.ll b/test/CodeGen/ARM64/indexed-vector-ldst.ll
index c909a447e1e15b4143ddbe5103867f1521044694..4e951f9d2f72ea84c37807d5c699aee062070300 100644 (file)
--- a/test/CodeGen/ARM64/indexed-vector-ldst.ll
+++ b/test/CodeGen/ARM64/indexed-vector-ldst.ll
@@ -611,3 +611,5080 @@ define float* @test_v2f32_post_reg_st1_lane(<2 x float> %in, float* %addr) {
   %newaddr = getelementptr float* %addr, i32 2
   ret float* %newaddr
 }
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld2(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld2:
+;CHECK: ld2.16b { v0, v1 }, [x0], #32
+  %ld2 = tail call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld2(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld2:
+;CHECK: ld2.16b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+declare { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld2(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld2:
+;CHECK: ld2.8b { v0, v1 }, [x0], #16
+  %ld2 = tail call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 16
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld2(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld2:
+;CHECK: ld2.8b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+declare { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld2(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld2:
+;CHECK: ld2.8h { v0, v1 }, [x0], #32
+  %ld2 = tail call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld2(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld2:
+;CHECK: ld2.8h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+declare { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld2(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld2:
+;CHECK: ld2.4h { v0, v1 }, [x0], #16
+  %ld2 = tail call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 8
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld2(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld2:
+;CHECK: ld2.4h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+declare { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld2(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld2:
+;CHECK: ld2.4s { v0, v1 }, [x0], #32
+  %ld2 = tail call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld2(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld2:
+;CHECK: ld2.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+declare { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld2(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld2:
+;CHECK: ld2.2s { v0, v1 }, [x0], #16
+  %ld2 = tail call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld2(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld2:
+;CHECK: ld2.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+declare { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld2(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld2:
+;CHECK: ld2.2d { v0, v1 }, [x0], #32
+  %ld2 = tail call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld2(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld2:
+;CHECK: ld2.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+declare { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld2(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld2:
+;CHECK: ld1.1d { v0, v1 }, [x0], #16
+  %ld2 = tail call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld2(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld2:
+;CHECK: ld1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+declare { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_imm_ld2(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld2:
+;CHECK: ld2.4s { v0, v1 }, [x0], #32
+  %ld2 = tail call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_reg_ld2(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld2:
+;CHECK: ld2.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+declare { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_imm_ld2(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld2:
+;CHECK: ld2.2s { v0, v1 }, [x0], #16
+  %ld2 = tail call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_reg_ld2(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld2:
+;CHECK: ld2.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+declare { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_imm_ld2(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld2:
+;CHECK: ld2.2d { v0, v1 }, [x0], #32
+  %ld2 = tail call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_reg_ld2(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld2:
+;CHECK: ld2.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+declare { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_imm_ld2(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld2:
+;CHECK: ld1.1d { v0, v1 }, [x0], #16
+  %ld2 = tail call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_reg_ld2(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld2:
+;CHECK: ld1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = tail call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+declare { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2.v1f64.p0f64(double*)
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld3(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld3:
+;CHECK: ld3.16b { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 48
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld3(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld3:
+;CHECK: ld3.16b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld3(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld3:
+;CHECK: ld3.8b { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 24
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld3(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld3:
+;CHECK: ld3.8b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld3(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld3:
+;CHECK: ld3.8h { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 24
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld3(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld3:
+;CHECK: ld3.8h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld3(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld3:
+;CHECK: ld3.4h { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 12
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld3(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld3:
+;CHECK: ld3.4h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld3(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld3:
+;CHECK: ld3.4s { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 12
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld3(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld3:
+;CHECK: ld3.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld3(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld3:
+;CHECK: ld3.2s { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 6
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld3(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld3:
+;CHECK: ld3.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld3(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld3:
+;CHECK: ld3.2d { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 6
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld3(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld3:
+;CHECK: ld3.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld3(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld3(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld3(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld3:
+;CHECK: ld3.4s { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 12
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld3(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld3:
+;CHECK: ld3.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+declare { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld3(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld3:
+;CHECK: ld3.2s { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 6
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld3(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld3:
+;CHECK: ld3.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+declare { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld3(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld3:
+;CHECK: ld3.2d { v0, v1, v2 }, [x0], #48
+  %ld3 = tail call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 6
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld3(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld3:
+;CHECK: ld3.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+declare { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld3(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], #24
+  %ld3 = tail call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld3(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = tail call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+declare { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3.v1f64.p0f64(double*)
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld4(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld4:
+;CHECK: ld4.16b { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 64
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld4(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld4:
+;CHECK: ld4.16b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld4(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld4:
+;CHECK: ld4.8b { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld4(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld4:
+;CHECK: ld4.8b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld4(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld4:
+;CHECK: ld4.8h { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 32
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld4(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld4:
+;CHECK: ld4.8h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld4(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld4:
+;CHECK: ld4.4h { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld4(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld4:
+;CHECK: ld4.4h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld4(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld4:
+;CHECK: ld4.4s { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 16
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld4(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld4:
+;CHECK: ld4.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld4(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld4:
+;CHECK: ld4.2s { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld4(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld4:
+;CHECK: ld4.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld4(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld4:
+;CHECK: ld4.2d { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 8
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld4(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld4:
+;CHECK: ld4.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld4(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld4(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld4(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld4:
+;CHECK: ld4.4s { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 16
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld4(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld4:
+;CHECK: ld4.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+declare { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld4(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld4:
+;CHECK: ld4.2s { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld4(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld4:
+;CHECK: ld4.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+declare { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld4(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld4:
+;CHECK: ld4.2d { v0, v1, v2, v3 }, [x0], #64
+  %ld4 = tail call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 8
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld4(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld4:
+;CHECK: ld4.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+declare { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld4(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = tail call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld4(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = tail call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+declare { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4.v1f64.p0f64(double*)
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld1x2(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld1x2:
+;CHECK: ld1.16b { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x2.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld1x2
+}
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld1x2(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld1x2:
+;CHECK: ld1.16b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x2.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld1x2
+}
+
+declare { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x2.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld1x2(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld1x2:
+;CHECK: ld1.8b { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x2.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 16
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld1x2
+}
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld1x2(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld1x2:
+;CHECK: ld1.8b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x2.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld1x2
+}
+
+declare { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x2.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld1x2(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld1x2:
+;CHECK: ld1.8h { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x2.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld1x2
+}
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld1x2(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld1x2:
+;CHECK: ld1.8h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x2.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld1x2
+}
+
+declare { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x2.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld1x2(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld1x2:
+;CHECK: ld1.4h { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x2.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 8
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld1x2
+}
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld1x2(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld1x2:
+;CHECK: ld1.4h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x2.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld1x2
+}
+
+declare { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x2.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld1x2(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld1x2:
+;CHECK: ld1.4s { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x2.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld1x2
+}
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld1x2(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld1x2:
+;CHECK: ld1.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x2.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld1x2
+}
+
+declare { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x2.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld1x2(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld1x2:
+;CHECK: ld1.2s { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x2.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld1x2
+}
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld1x2(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld1x2:
+;CHECK: ld1.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x2.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld1x2
+}
+
+declare { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x2.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld1x2(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld1x2:
+;CHECK: ld1.2d { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x2.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld1x2
+}
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld1x2(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld1x2:
+;CHECK: ld1.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x2.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld1x2
+}
+
+declare { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x2.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld1x2(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld1x2:
+;CHECK: ld1.1d { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x2.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld1x2
+}
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld1x2(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld1x2:
+;CHECK: ld1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x2.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld1x2
+}
+
+declare { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x2.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_imm_ld1x2(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld1x2:
+;CHECK: ld1.4s { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x2.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld1x2
+}
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_reg_ld1x2(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld1x2:
+;CHECK: ld1.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x2.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld1x2
+}
+
+declare { <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x2.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_imm_ld1x2(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld1x2:
+;CHECK: ld1.2s { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x2.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld1x2
+}
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_reg_ld1x2(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld1x2:
+;CHECK: ld1.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x2.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld1x2
+}
+
+declare { <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x2.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_imm_ld1x2(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld1x2:
+;CHECK: ld1.2d { v0, v1 }, [x0], #32
+  %ld1x2 = tail call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x2.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld1x2
+}
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_reg_ld1x2(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld1x2:
+;CHECK: ld1.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x2.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld1x2
+}
+
+declare { <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x2.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_imm_ld1x2(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld1x2:
+;CHECK: ld1.1d { v0, v1 }, [x0], #16
+  %ld1x2 = tail call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x2.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld1x2
+}
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_reg_ld1x2(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld1x2:
+;CHECK: ld1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld1x2 = tail call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x2.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld1x2
+}
+
+declare { <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x2.v1f64.p0f64(double*)
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld1x3(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld1x3:
+;CHECK: ld1.16b { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x3.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 48
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld1x3
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld1x3(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld1x3:
+;CHECK: ld1.16b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x3.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld1x3
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x3.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld1x3(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld1x3:
+;CHECK: ld1.8b { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x3.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 24
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld1x3
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld1x3(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld1x3:
+;CHECK: ld1.8b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x3.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld1x3
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x3.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld1x3(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld1x3:
+;CHECK: ld1.8h { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x3.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 24
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld1x3
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld1x3(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld1x3:
+;CHECK: ld1.8h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x3.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld1x3
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x3.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld1x3(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld1x3:
+;CHECK: ld1.4h { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x3.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 12
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld1x3
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld1x3(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld1x3:
+;CHECK: ld1.4h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x3.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld1x3
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x3.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld1x3(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld1x3:
+;CHECK: ld1.4s { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x3.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 12
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld1x3
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld1x3(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld1x3:
+;CHECK: ld1.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x3.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld1x3
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x3.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld1x3(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld1x3:
+;CHECK: ld1.2s { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x3.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 6
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld1x3
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld1x3(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld1x3:
+;CHECK: ld1.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x3.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld1x3
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x3.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld1x3(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld1x3:
+;CHECK: ld1.2d { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x3.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 6
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld1x3
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld1x3(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld1x3:
+;CHECK: ld1.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x3.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld1x3
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x3.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld1x3(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld1x3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x3.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld1x3
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld1x3(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld1x3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x3.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld1x3
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x3.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld1x3(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld1x3:
+;CHECK: ld1.4s { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x3.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 12
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld1x3
+}
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld1x3(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld1x3:
+;CHECK: ld1.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x3.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld1x3
+}
+
+declare { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x3.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld1x3(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld1x3:
+;CHECK: ld1.2s { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x3.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 6
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld1x3
+}
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld1x3(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld1x3:
+;CHECK: ld1.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x3.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld1x3
+}
+
+declare { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x3.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld1x3(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld1x3:
+;CHECK: ld1.2d { v0, v1, v2 }, [x0], #48
+  %ld1x3 = tail call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x3.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 6
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld1x3
+}
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld1x3(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld1x3:
+;CHECK: ld1.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x3.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld1x3
+}
+
+declare { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x3.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld1x3(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld1x3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], #24
+  %ld1x3 = tail call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x3.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld1x3
+}
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld1x3(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld1x3:
+;CHECK: ld1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld1x3 = tail call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x3.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld1x3
+}
+
+declare { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x3.v1f64.p0f64(double*)
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld1x4(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v16i8_post_imm_ld1x4:
+;CHECK: ld1.16b { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x4.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 64
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld1x4
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld1x4(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v16i8_post_reg_ld1x4:
+;CHECK: ld1.16b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x4.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld1x4
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld1x4.v16i8.p0i8(i8*)
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld1x4(i8* %A, i8** %ptr) {
+;CHECK-LABEL: test_v8i8_post_imm_ld1x4:
+;CHECK: ld1.8b { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x4.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld1x4
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld1x4(i8* %A, i8** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i8_post_reg_ld1x4:
+;CHECK: ld1.8b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x4.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld1x4
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld1x4.v8i8.p0i8(i8*)
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld1x4(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v8i16_post_imm_ld1x4:
+;CHECK: ld1.8h { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x4.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 32
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld1x4
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld1x4(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v8i16_post_reg_ld1x4:
+;CHECK: ld1.8h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x4.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld1x4
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld1x4.v8i16.p0i16(i16*)
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld1x4(i16* %A, i16** %ptr) {
+;CHECK-LABEL: test_v4i16_post_imm_ld1x4:
+;CHECK: ld1.4h { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x4.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld1x4
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld1x4(i16* %A, i16** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i16_post_reg_ld1x4:
+;CHECK: ld1.4h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x4.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld1x4
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld1x4.v4i16.p0i16(i16*)
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld1x4(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v4i32_post_imm_ld1x4:
+;CHECK: ld1.4s { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x4.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 16
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld1x4
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld1x4(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4i32_post_reg_ld1x4:
+;CHECK: ld1.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x4.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld1x4
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld1x4.v4i32.p0i32(i32*)
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld1x4(i32* %A, i32** %ptr) {
+;CHECK-LABEL: test_v2i32_post_imm_ld1x4:
+;CHECK: ld1.2s { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x4.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld1x4
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld1x4(i32* %A, i32** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i32_post_reg_ld1x4:
+;CHECK: ld1.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x4.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld1x4
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld1x4.v2i32.p0i32(i32*)
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld1x4(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v2i64_post_imm_ld1x4:
+;CHECK: ld1.2d { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x4.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 8
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld1x4
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld1x4(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2i64_post_reg_ld1x4:
+;CHECK: ld1.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x4.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld1x4
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld1x4.v2i64.p0i64(i64*)
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld1x4(i64* %A, i64** %ptr) {
+;CHECK-LABEL: test_v1i64_post_imm_ld1x4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x4.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld1x4
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld1x4(i64* %A, i64** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1i64_post_reg_ld1x4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x4.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld1x4
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld1x4.v1i64.p0i64(i64*)
+
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld1x4(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v4f32_post_imm_ld1x4:
+;CHECK: ld1.4s { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x4.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 16
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld1x4
+}
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld1x4(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v4f32_post_reg_ld1x4:
+;CHECK: ld1.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x4.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld1x4
+}
+
+declare { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld1x4.v4f32.p0f32(float*)
+
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld1x4(float* %A, float** %ptr) {
+;CHECK-LABEL: test_v2f32_post_imm_ld1x4:
+;CHECK: ld1.2s { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x4.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld1x4
+}
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld1x4(float* %A, float** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f32_post_reg_ld1x4:
+;CHECK: ld1.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x4.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld1x4
+}
+
+declare { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld1x4.v2f32.p0f32(float*)
+
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld1x4(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v2f64_post_imm_ld1x4:
+;CHECK: ld1.2d { v0, v1, v2, v3 }, [x0], #64
+  %ld1x4 = tail call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x4.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 8
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld1x4
+}
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld1x4(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v2f64_post_reg_ld1x4:
+;CHECK: ld1.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x4.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld1x4
+}
+
+declare { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld1x4.v2f64.p0f64(double*)
+
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld1x4(double* %A, double** %ptr) {
+;CHECK-LABEL: test_v1f64_post_imm_ld1x4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld1x4 = tail call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x4.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld1x4
+}
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld1x4(double* %A, double** %ptr, i64 %inc) {
+;CHECK-LABEL: test_v1f64_post_reg_ld1x4:
+;CHECK: ld1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld1x4 = tail call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x4.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld1x4
+}
+
+declare { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld1x4.v1f64.p0f64(double*)
+
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld2r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld2r:
+;CHECK: ld2r.16b { v0, v1 }, [x0], #2
+  %ld2 = call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld2r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld2r:
+;CHECK: ld2r.16b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+declare { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2r.v16i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld2r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld2r:
+;CHECK: ld2r.8b { v0, v1 }, [x0], #2
+  %ld2 = call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld2r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld2r:
+;CHECK: ld2r.8b { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+declare { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2r.v8i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld2r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld2r:
+;CHECK: ld2r.8h { v0, v1 }, [x0], #4
+  %ld2 = call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld2r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld2r:
+;CHECK: ld2r.8h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+declare { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2r.v8i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld2r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld2r:
+;CHECK: ld2r.4h { v0, v1 }, [x0], #4
+  %ld2 = call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld2r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld2r:
+;CHECK: ld2r.4h { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+declare { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2r.v4i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld2r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld2r:
+;CHECK: ld2r.4s { v0, v1 }, [x0], #8
+  %ld2 = call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld2r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld2r:
+;CHECK: ld2r.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+declare { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2r.v4i32.p0i32(i32*) nounwind readonly
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld2r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld2r:
+;CHECK: ld2r.2s { v0, v1 }, [x0], #8
+  %ld2 = call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld2r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld2r:
+;CHECK: ld2r.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+declare { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2r.v2i32.p0i32(i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld2r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld2r:
+;CHECK: ld2r.2d { v0, v1 }, [x0], #16
+  %ld2 = call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld2r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld2r:
+;CHECK: ld2r.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+declare { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2r.v2i64.p0i64(i64*) nounwind readonly
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld2r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld2r:
+;CHECK: ld2r.1d { v0, v1 }, [x0], #16
+  %ld2 = call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld2r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld2r:
+;CHECK: ld2r.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+declare { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2r.v1i64.p0i64(i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_imm_ld2r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld2r:
+;CHECK: ld2r.4s { v0, v1 }, [x0], #8
+  %ld2 = call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_reg_ld2r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld2r:
+;CHECK: ld2r.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+declare { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2r.v4f32.p0f32(float*) nounwind readonly
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_imm_ld2r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld2r:
+;CHECK: ld2r.2s { v0, v1 }, [x0], #8
+  %ld2 = call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_reg_ld2r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld2r:
+;CHECK: ld2r.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+declare { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2r.v2f32.p0f32(float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_imm_ld2r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld2r:
+;CHECK: ld2r.2d { v0, v1 }, [x0], #16
+  %ld2 = call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_reg_ld2r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld2r:
+;CHECK: ld2r.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+declare { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2r.v2f64.p0f64(double*) nounwind readonly
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_imm_ld2r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld2r:
+;CHECK: ld2r.1d { v0, v1 }, [x0], #16
+  %ld2 = call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_reg_ld2r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld2r:
+;CHECK: ld2r.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  %ld2 = call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+declare { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2r.v1f64.p0f64(double*) nounwind readonly
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld3r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld3r:
+;CHECK: ld3r.16b { v0, v1, v2 }, [x0], #3
+  %ld3 = call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld3r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld3r:
+;CHECK: ld3r.16b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3r.v16i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld3r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld3r:
+;CHECK: ld3r.8b { v0, v1, v2 }, [x0], #3
+  %ld3 = call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld3r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld3r:
+;CHECK: ld3r.8b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3r.v8i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld3r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld3r:
+;CHECK: ld3r.8h { v0, v1, v2 }, [x0], #6
+  %ld3 = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld3r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld3r:
+;CHECK: ld3r.8h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3r.v8i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld3r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld3r:
+;CHECK: ld3r.4h { v0, v1, v2 }, [x0], #6
+  %ld3 = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld3r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld3r:
+;CHECK: ld3r.4h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3r.v4i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld3r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld3r:
+;CHECK: ld3r.4s { v0, v1, v2 }, [x0], #12
+  %ld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld3r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld3r:
+;CHECK: ld3r.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3r.v4i32.p0i32(i32*) nounwind readonly
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld3r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld3r:
+;CHECK: ld3r.2s { v0, v1, v2 }, [x0], #12
+  %ld3 = call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld3r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld3r:
+;CHECK: ld3r.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3r.v2i32.p0i32(i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld3r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld3r:
+;CHECK: ld3r.2d { v0, v1, v2 }, [x0], #24
+  %ld3 = call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld3r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld3r:
+;CHECK: ld3r.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3r.v2i64.p0i64(i64*) nounwind readonly
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld3r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld3r:
+;CHECK: ld3r.1d { v0, v1, v2 }, [x0], #24
+  %ld3 = call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld3r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld3r:
+;CHECK: ld3r.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3r.v1i64.p0i64(i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld3r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld3r:
+;CHECK: ld3r.4s { v0, v1, v2 }, [x0], #12
+  %ld3 = call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld3r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld3r:
+;CHECK: ld3r.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+declare { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3r.v4f32.p0f32(float*) nounwind readonly
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld3r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld3r:
+;CHECK: ld3r.2s { v0, v1, v2 }, [x0], #12
+  %ld3 = call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld3r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld3r:
+;CHECK: ld3r.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+declare { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3r.v2f32.p0f32(float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld3r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld3r:
+;CHECK: ld3r.2d { v0, v1, v2 }, [x0], #24
+  %ld3 = call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld3r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld3r:
+;CHECK: ld3r.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+declare { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3r.v2f64.p0f64(double*) nounwind readonly
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld3r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld3r:
+;CHECK: ld3r.1d { v0, v1, v2 }, [x0], #24
+  %ld3 = call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld3r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld3r:
+;CHECK: ld3r.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  %ld3 = call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+declare { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3r.v1f64.p0f64(double*) nounwind readonly
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld4r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld4r:
+;CHECK: ld4r.16b { v0, v1, v2, v3 }, [x0], #4
+  %ld4 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld4r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld4r:
+;CHECK: ld4r.16b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4r.v16i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4r.v16i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld4r(i8* %A, i8** %ptr) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld4r:
+;CHECK: ld4r.8b { v0, v1, v2, v3 }, [x0], #4
+  %ld4 = call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld4r(i8* %A, i8** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld4r:
+;CHECK: ld4r.8b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4r.v8i8.p0i8(i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4r.v8i8.p0i8(i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld4r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld4r:
+;CHECK: ld4r.8h { v0, v1, v2, v3 }, [x0], #8
+  %ld4 = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld4r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld4r:
+;CHECK: ld4r.8h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4r.v8i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4r.v8i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld4r(i16* %A, i16** %ptr) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld4r:
+;CHECK: ld4r.4h { v0, v1, v2, v3 }, [x0], #8
+  %ld4 = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld4r(i16* %A, i16** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld4r:
+;CHECK: ld4r.4h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4r.v4i16.p0i16(i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4r.v4i16.p0i16(i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld4r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld4r:
+;CHECK: ld4r.4s { v0, v1, v2, v3 }, [x0], #16
+  %ld4 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld4r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld4r:
+;CHECK: ld4r.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4r.v4i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4r.v4i32.p0i32(i32*) nounwind readonly
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld4r(i32* %A, i32** %ptr) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld4r:
+;CHECK: ld4r.2s { v0, v1, v2, v3 }, [x0], #16
+  %ld4 = call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld4r(i32* %A, i32** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld4r:
+;CHECK: ld4r.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4r.v2i32.p0i32(i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4r.v2i32.p0i32(i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld4r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld4r:
+;CHECK: ld4r.2d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld4r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld4r:
+;CHECK: ld4r.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4r.v2i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4r.v2i64.p0i64(i64*) nounwind readonly
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld4r(i64* %A, i64** %ptr) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld4r:
+;CHECK: ld4r.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld4r(i64* %A, i64** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld4r:
+;CHECK: ld4r.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4r.v1i64.p0i64(i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4r.v1i64.p0i64(i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld4r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld4r:
+;CHECK: ld4r.4s { v0, v1, v2, v3 }, [x0], #16
+  %ld4 = call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld4r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld4r:
+;CHECK: ld4r.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4r.v4f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+declare { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4r.v4f32.p0f32(float*) nounwind readonly
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld4r(float* %A, float** %ptr) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld4r:
+;CHECK: ld4r.2s { v0, v1, v2, v3 }, [x0], #16
+  %ld4 = call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld4r(float* %A, float** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld4r:
+;CHECK: ld4r.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4r.v2f32.p0f32(float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+declare { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4r.v2f32.p0f32(float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld4r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld4r:
+;CHECK: ld4r.2d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld4r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld4r:
+;CHECK: ld4r.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4r.v2f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+declare { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4r.v2f64.p0f64(double*) nounwind readonly
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld4r(double* %A, double** %ptr) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld4r:
+;CHECK: ld4r.1d { v0, v1, v2, v3 }, [x0], #32
+  %ld4 = call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld4r(double* %A, double** %ptr, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld4r:
+;CHECK: ld4r.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  %ld4 = call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4r.v1f64.p0f64(double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+declare { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4r.v1f64.p0f64(double*) nounwind readonly
+
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld2lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld2lane:
+;CHECK: ld2.b { v0, v1 }[0], [x0], #2
+  %ld2 = call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+define { <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld2lane(i8* %A, i8** %ptr, i64 %inc, <16 x i8> %B, <16 x i8> %C) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld2lane:
+;CHECK: ld2.b { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8> } %ld2
+}
+
+declare { <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld2lane.v16i8.p0i8(<16 x i8>, <16 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld2lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld2lane:
+;CHECK: ld2.b { v0, v1 }[0], [x0], #2
+  %ld2 = call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+define { <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld2lane(i8* %A, i8** %ptr, i64 %inc, <8 x i8> %B, <8 x i8> %C) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld2lane:
+;CHECK: ld2.b { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8> } %ld2
+}
+
+declare { <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld2lane.v8i8.p0i8(<8 x i8>, <8 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld2lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld2lane:
+;CHECK: ld2.h { v0, v1 }[0], [x0], #4
+  %ld2 = call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+define { <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld2lane(i16* %A, i16** %ptr, i64 %inc, <8 x i16> %B, <8 x i16> %C) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld2lane:
+;CHECK: ld2.h { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16> } %ld2
+}
+
+declare { <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld2lane.v8i16.p0i16(<8 x i16>, <8 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld2lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld2lane:
+;CHECK: ld2.h { v0, v1 }[0], [x0], #4
+  %ld2 = call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+define { <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld2lane(i16* %A, i16** %ptr, i64 %inc, <4 x i16> %B, <4 x i16> %C) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld2lane:
+;CHECK: ld2.h { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16> } %ld2
+}
+
+declare { <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld2lane.v4i16.p0i16(<4 x i16>, <4 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld2lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], #8
+  %ld2 = call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+define { <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld2lane(i32* %A, i32** %ptr, i64 %inc, <4 x i32> %B, <4 x i32> %C) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32> } %ld2
+}
+
+declare { <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld2lane.v4i32.p0i32(<4 x i32>, <4 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld2lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], #8
+  %ld2 = call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+define { <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld2lane(i32* %A, i32** %ptr, i64 %inc, <2 x i32> %B, <2 x i32> %C) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32> } %ld2
+}
+
+declare { <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld2lane.v2i32.p0i32(<2 x i32>, <2 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld2lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], #16
+  %ld2 = call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+define { <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld2lane(i64* %A, i64** %ptr, i64 %inc, <2 x i64> %B, <2 x i64> %C) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64> } %ld2
+}
+
+declare { <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld2lane.v2i64.p0i64(<2 x i64>, <2 x i64>, i64, i64*) nounwind readonly
+
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld2lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], #16
+  %ld2 = call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 2
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+define { <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld2lane(i64* %A, i64** %ptr, i64 %inc, <1 x i64> %B, <1 x i64> %C) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64> } %ld2
+}
+
+declare { <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld2lane.v1i64.p0i64(<1 x i64>, <1 x i64>, i64, i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_imm_ld2lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], #8
+  %ld2 = call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+define { <4 x float>, <4 x float> } @test_v4f32_post_reg_ld2lane(float* %A, float** %ptr, i64 %inc, <4 x float> %B, <4 x float> %C) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float> } %ld2
+}
+
+declare { <4 x float>, <4 x float> } @llvm.arm64.neon.ld2lane.v4f32.p0f32(<4 x float>, <4 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_imm_ld2lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], #8
+  %ld2 = call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+define { <2 x float>, <2 x float> } @test_v2f32_post_reg_ld2lane(float* %A, float** %ptr, i64 %inc, <2 x float> %B, <2 x float> %C) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld2lane:
+;CHECK: ld2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float> } %ld2
+}
+
+declare { <2 x float>, <2 x float> } @llvm.arm64.neon.ld2lane.v2f32.p0f32(<2 x float>, <2 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_imm_ld2lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], #16
+  %ld2 = call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+define { <2 x double>, <2 x double> } @test_v2f64_post_reg_ld2lane(double* %A, double** %ptr, i64 %inc, <2 x double> %B, <2 x double> %C) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double> } %ld2
+}
+
+declare { <2 x double>, <2 x double> } @llvm.arm64.neon.ld2lane.v2f64.p0f64(<2 x double>, <2 x double>, i64, double*) nounwind readonly
+
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_imm_ld2lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], #16
+  %ld2 = call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 2
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+define { <1 x double>, <1 x double> } @test_v1f64_post_reg_ld2lane(double* %A, double** %ptr, i64 %inc, <1 x double> %B, <1 x double> %C) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld2lane:
+;CHECK: ld2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  %ld2 = call { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double> } %ld2
+}
+
+declare { <1 x double>, <1 x double> } @llvm.arm64.neon.ld2lane.v1f64.p0f64(<1 x double>, <1 x double>, i64, double*) nounwind readonly
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld3lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld3lane:
+;CHECK: ld3.b { v0, v1, v2 }[0], [x0], #3
+  %ld3 = call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld3lane(i8* %A, i8** %ptr, i64 %inc, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld3lane:
+;CHECK: ld3.b { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8> } %ld3
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld3lane.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld3lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld3lane:
+;CHECK: ld3.b { v0, v1, v2 }[0], [x0], #3
+  %ld3 = call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld3lane(i8* %A, i8** %ptr, i64 %inc, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld3lane:
+;CHECK: ld3.b { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8> } %ld3
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld3lane.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld3lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld3lane:
+;CHECK: ld3.h { v0, v1, v2 }[0], [x0], #6
+  %ld3 = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld3lane(i16* %A, i16** %ptr, i64 %inc, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld3lane:
+;CHECK: ld3.h { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16> } %ld3
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld3lane.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld3lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld3lane:
+;CHECK: ld3.h { v0, v1, v2 }[0], [x0], #6
+  %ld3 = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld3lane(i16* %A, i16** %ptr, i64 %inc, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld3lane:
+;CHECK: ld3.h { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16> } %ld3
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld3lane.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld3lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], #12
+  %ld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld3lane(i32* %A, i32** %ptr, i64 %inc, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32> } %ld3
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld3lane.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld3lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], #12
+  %ld3 = call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld3lane(i32* %A, i32** %ptr, i64 %inc, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32> } %ld3
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld3lane.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld3lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], #24
+  %ld3 = call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld3lane(i64* %A, i64** %ptr, i64 %inc, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64> } %ld3
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld3lane.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, i64, i64*) nounwind readonly
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld3lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], #24
+  %ld3 = call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 3
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld3lane(i64* %A, i64** %ptr, i64 %inc, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64> } %ld3
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld3lane.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, i64, i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld3lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], #12
+  %ld3 = call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+define { <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld3lane(float* %A, float** %ptr, i64 %inc, <4 x float> %B, <4 x float> %C, <4 x float> %D) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float> } %ld3
+}
+
+declare { <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld3lane.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld3lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], #12
+  %ld3 = call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+define { <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld3lane(float* %A, float** %ptr, i64 %inc, <2 x float> %B, <2 x float> %C, <2 x float> %D) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld3lane:
+;CHECK: ld3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float> } %ld3
+}
+
+declare { <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld3lane.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld3lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], #24
+  %ld3 = call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+define { <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld3lane(double* %A, double** %ptr, i64 %inc, <2 x double> %B, <2 x double> %C, <2 x double> %D) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double> } %ld3
+}
+
+declare { <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld3lane.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, i64, double*) nounwind readonly
+
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld3lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], #24
+  %ld3 = call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 3
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+define { <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld3lane(double* %A, double** %ptr, i64 %inc, <1 x double> %B, <1 x double> %C, <1 x double> %D) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld3lane:
+;CHECK: ld3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  %ld3 = call { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double> } %ld3
+}
+
+declare { <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld3lane.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, i64, double*) nounwind readonly
+
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_imm_ld4lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_ld4lane:
+;CHECK: ld4.b { v0, v1, v2, v3 }[0], [x0], #4
+  %ld4 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+define { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @test_v16i8_post_reg_ld4lane(i8* %A, i8** %ptr, i64 %inc, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_ld4lane:
+;CHECK: ld4.b { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } %ld4
+}
+
+declare { <16 x i8>, <16 x i8>, <16 x i8>, <16 x i8> } @llvm.arm64.neon.ld4lane.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_imm_ld4lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_ld4lane:
+;CHECK: ld4.b { v0, v1, v2, v3 }[0], [x0], #4
+  %ld4 = call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+define { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @test_v8i8_post_reg_ld4lane(i8* %A, i8** %ptr, i64 %inc, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_ld4lane:
+;CHECK: ld4.b { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  store i8* %tmp, i8** %ptr
+  ret { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } %ld4
+}
+
+declare { <8 x i8>, <8 x i8>, <8 x i8>, <8 x i8> } @llvm.arm64.neon.ld4lane.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>, i64, i8*) nounwind readonly
+
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_imm_ld4lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_ld4lane:
+;CHECK: ld4.h { v0, v1, v2, v3 }[0], [x0], #8
+  %ld4 = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+define { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @test_v8i16_post_reg_ld4lane(i16* %A, i16** %ptr, i64 %inc, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_ld4lane:
+;CHECK: ld4.h { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } %ld4
+}
+
+declare { <8 x i16>, <8 x i16>, <8 x i16>, <8 x i16> } @llvm.arm64.neon.ld4lane.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_imm_ld4lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_ld4lane:
+;CHECK: ld4.h { v0, v1, v2, v3 }[0], [x0], #8
+  %ld4 = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+define { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @test_v4i16_post_reg_ld4lane(i16* %A, i16** %ptr, i64 %inc, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_ld4lane:
+;CHECK: ld4.h { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  store i16* %tmp, i16** %ptr
+  ret { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } %ld4
+}
+
+declare { <4 x i16>, <4 x i16>, <4 x i16>, <4 x i16> } @llvm.arm64.neon.ld4lane.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>, i64, i16*) nounwind readonly
+
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_imm_ld4lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], #16
+  %ld4 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+define { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @test_v4i32_post_reg_ld4lane(i32* %A, i32** %ptr, i64 %inc, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } %ld4
+}
+
+declare { <4 x i32>, <4 x i32>, <4 x i32>, <4 x i32> } @llvm.arm64.neon.ld4lane.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_imm_ld4lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], #16
+  %ld4 = call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+define { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @test_v2i32_post_reg_ld4lane(i32* %A, i32** %ptr, i64 %inc, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  store i32* %tmp, i32** %ptr
+  ret { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } %ld4
+}
+
+declare { <2 x i32>, <2 x i32>, <2 x i32>, <2 x i32> } @llvm.arm64.neon.ld4lane.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, i64, i32*) nounwind readonly
+
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_imm_ld4lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], #32
+  %ld4 = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+define { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @test_v2i64_post_reg_ld4lane(i64* %A, i64** %ptr, i64 %inc, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } %ld4
+}
+
+declare { <2 x i64>, <2 x i64>, <2 x i64>, <2 x i64> } @llvm.arm64.neon.ld4lane.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, <2 x i64>, i64, i64*) nounwind readonly
+
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_imm_ld4lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], #32
+  %ld4 = call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i32 4
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+define { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @test_v1i64_post_reg_ld4lane(i64* %A, i64** %ptr, i64 %inc, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  store i64* %tmp, i64** %ptr
+  ret { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } %ld4
+}
+
+declare { <1 x i64>, <1 x i64>, <1 x i64>, <1 x i64> } @llvm.arm64.neon.ld4lane.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, <1 x i64>, i64, i64*) nounwind readonly
+
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_imm_ld4lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], #16
+  %ld4 = call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+define { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @test_v4f32_post_reg_ld4lane(float* %A, float** %ptr, i64 %inc, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <4 x float>, <4 x float>, <4 x float>, <4 x float> } %ld4
+}
+
+declare { <4 x float>, <4 x float>, <4 x float>, <4 x float> } @llvm.arm64.neon.ld4lane.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, <4 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_imm_ld4lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], #16
+  %ld4 = call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+define { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @test_v2f32_post_reg_ld4lane(float* %A, float** %ptr, i64 %inc, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_ld4lane:
+;CHECK: ld4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  store float* %tmp, float** %ptr
+  ret { <2 x float>, <2 x float>, <2 x float>, <2 x float> } %ld4
+}
+
+declare { <2 x float>, <2 x float>, <2 x float>, <2 x float> } @llvm.arm64.neon.ld4lane.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, <2 x float>, i64, float*) nounwind readonly
+
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_imm_ld4lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], #32
+  %ld4 = call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+define { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @test_v2f64_post_reg_ld4lane(double* %A, double** %ptr, i64 %inc, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <2 x double>, <2 x double>, <2 x double>, <2 x double> } %ld4
+}
+
+declare { <2 x double>, <2 x double>, <2 x double>, <2 x double> } @llvm.arm64.neon.ld4lane.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, <2 x double>, i64, double*) nounwind readonly
+
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_imm_ld4lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], #32
+  %ld4 = call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i32 4
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+define { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @test_v1f64_post_reg_ld4lane(double* %A, double** %ptr, i64 %inc, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_ld4lane:
+;CHECK: ld4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  %ld4 = call { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  store double* %tmp, double** %ptr
+  ret { <1 x double>, <1 x double>, <1 x double>, <1 x double> } %ld4
+}
+
+declare { <1 x double>, <1 x double>, <1 x double>, <1 x double> } @llvm.arm64.neon.ld4lane.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, <1 x double>, i64, double*) nounwind readonly
+
+
+define i8* @test_v16i8_post_imm_st2(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st2:
+;CHECK: st2.16b { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st2(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st2:
+;CHECK: st2.16b { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v16i8.p0i8(<16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st2(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st2:
+;CHECK: st2.8b { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i32 16
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st2(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st2:
+;CHECK: st2.8b { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v8i8.p0i8(<8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st2(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st2:
+;CHECK: st2.8h { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st2(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st2:
+;CHECK: st2.8h { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v8i16.p0i16(<8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st2(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st2:
+;CHECK: st2.4h { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i32 8
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st2(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st2:
+;CHECK: st2.4h { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v4i16.p0i16(<4 x i16>, <4 x i16>, i16*)
+
+
+define i32* @test_v4i32_post_imm_st2(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st2:
+;CHECK: st2.4s { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st2(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st2:
+;CHECK: st2.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v4i32.p0i32(<4 x i32>, <4 x i32>, i32*)
+
+
+define i32* @test_v2i32_post_imm_st2(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st2:
+;CHECK: st2.2s { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st2(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st2:
+;CHECK: st2.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v2i32.p0i32(<2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st2(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st2:
+;CHECK: st2.2d { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st2(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st2:
+;CHECK: st2.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v2i64.p0i64(<2 x i64>, <2 x i64>, i64*)
+
+
+define i64* @test_v1i64_post_imm_st2(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st2:
+;CHECK: st1.1d { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 2
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st2(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st2:
+;CHECK: st1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v1i64.p0i64(<1 x i64>, <1 x i64>, i64*)
+
+
+define float* @test_v4f32_post_imm_st2(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st2:
+;CHECK: st2.4s { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v4f32.p0f32(<4 x float> %B, <4 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st2(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st2:
+;CHECK: st2.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v4f32.p0f32(<4 x float> %B, <4 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v4f32.p0f32(<4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st2(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st2:
+;CHECK: st2.2s { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v2f32.p0f32(<2 x float> %B, <2 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st2(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st2:
+;CHECK: st2.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v2f32.p0f32(<2 x float> %B, <2 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v2f32.p0f32(<2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st2(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st2:
+;CHECK: st2.2d { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st2.v2f64.p0f64(<2 x double> %B, <2 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st2(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st2:
+;CHECK: st2.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v2f64.p0f64(<2 x double> %B, <2 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v2f64.p0f64(<2 x double>, <2 x double>, double*)
+
+
+define double* @test_v1f64_post_imm_st2(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st2:
+;CHECK: st1.1d { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st2.v1f64.p0f64(<1 x double> %B, <1 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 2
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st2(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st2:
+;CHECK: st1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2.v1f64.p0f64(<1 x double> %B, <1 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st2.v1f64.p0f64(<1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st3(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st3:
+;CHECK: st3.16b { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i32 48
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st3(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st3:
+;CHECK: st3.16b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st3(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st3:
+;CHECK: st3.8b { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i32 24
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st3(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st3:
+;CHECK: st3.8b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st3(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st3:
+;CHECK: st3.8h { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i32 24
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st3(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st3:
+;CHECK: st3.8h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st3(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st3:
+;CHECK: st3.4h { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i32 12
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st3(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st3:
+;CHECK: st3.4h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, i16*)
+
+
+define i32* @test_v4i32_post_imm_st3(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st3:
+;CHECK: st3.4s { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i32 12
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st3(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st3:
+;CHECK: st3.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, i32*)
+
+
+define i32* @test_v2i32_post_imm_st3(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st3:
+;CHECK: st3.2s { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i32 6
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st3(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st3:
+;CHECK: st3.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st3(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st3:
+;CHECK: st3.2d { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 6
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st3(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st3:
+;CHECK: st3.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, i64*)
+
+
+define i64* @test_v1i64_post_imm_st3(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 3
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st3(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, i64*)
+
+
+define float* @test_v4f32_post_imm_st3(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st3:
+;CHECK: st3.4s { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i32 12
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st3(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st3:
+;CHECK: st3.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st3(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st3:
+;CHECK: st3.2s { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i32 6
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st3(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st3:
+;CHECK: st3.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st3(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st3:
+;CHECK: st3.2d { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st3.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 6
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st3(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st3:
+;CHECK: st3.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, double*)
+
+
+define double* @test_v1f64_post_imm_st3(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st3.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 3
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st3(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st3.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st4(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st4:
+;CHECK: st4.16b { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i32 64
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st4(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st4:
+;CHECK: st4.16b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st4(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st4:
+;CHECK: st4.8b { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st4(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st4:
+;CHECK: st4.8b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st4(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st4:
+;CHECK: st4.8h { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i32 32
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st4(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st4:
+;CHECK: st4.8h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st4(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st4:
+;CHECK: st4.4h { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st4(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st4:
+;CHECK: st4.4h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>,<4 x i16>,  i16*)
+
+
+define i32* @test_v4i32_post_imm_st4(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st4:
+;CHECK: st4.4s { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i32 16
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st4(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st4:
+;CHECK: st4.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>,<4 x i32>,  i32*)
+
+
+define i32* @test_v2i32_post_imm_st4(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st4:
+;CHECK: st4.2s { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st4(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st4:
+;CHECK: st4.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st4(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st4:
+;CHECK: st4.2d { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 8
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st4(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st4:
+;CHECK: st4.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>,<2 x i64>,  i64*)
+
+
+define i64* @test_v1i64_post_imm_st4(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st4(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>,<1 x i64>,  i64*)
+
+
+define float* @test_v4f32_post_imm_st4(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st4:
+;CHECK: st4.4s { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i32 16
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st4(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st4:
+;CHECK: st4.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st4(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st4:
+;CHECK: st4.2s { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st4(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st4:
+;CHECK: st4.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st4(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st4:
+;CHECK: st4.2d { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st4.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 8
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st4(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st4:
+;CHECK: st4.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>,<2 x double>,  double*)
+
+
+define double* @test_v1f64_post_imm_st4(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st4.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st4(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st4.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st1x2(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st1x2:
+;CHECK: st1.16b { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st1x2(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st1x2:
+;CHECK: st1.16b { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v16i8.p0i8(<16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st1x2(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st1x2:
+;CHECK: st1.8b { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i32 16
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st1x2(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st1x2:
+;CHECK: st1.8b { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v8i8.p0i8(<8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st1x2(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st1x2:
+;CHECK: st1.8h { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st1x2(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st1x2:
+;CHECK: st1.8h { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v8i16.p0i16(<8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st1x2(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st1x2:
+;CHECK: st1.4h { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i32 8
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st1x2(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st1x2:
+;CHECK: st1.4h { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v4i16.p0i16(<4 x i16>, <4 x i16>, i16*)
+
+
+define i32* @test_v4i32_post_imm_st1x2(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st1x2:
+;CHECK: st1.4s { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st1x2(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st1x2:
+;CHECK: st1.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v4i32.p0i32(<4 x i32>, <4 x i32>, i32*)
+
+
+define i32* @test_v2i32_post_imm_st1x2(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st1x2:
+;CHECK: st1.2s { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st1x2(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st1x2:
+;CHECK: st1.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v2i32.p0i32(<2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st1x2(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st1x2:
+;CHECK: st1.2d { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st1x2(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st1x2:
+;CHECK: st1.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v2i64.p0i64(<2 x i64>, <2 x i64>, i64*)
+
+
+define i64* @test_v1i64_post_imm_st1x2(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st1x2:
+;CHECK: st1.1d { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 2
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st1x2(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st1x2:
+;CHECK: st1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v1i64.p0i64(<1 x i64>, <1 x i64>, i64*)
+
+
+define float* @test_v4f32_post_imm_st1x2(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st1x2:
+;CHECK: st1.4s { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v4f32.p0f32(<4 x float> %B, <4 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st1x2(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st1x2:
+;CHECK: st1.4s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v4f32.p0f32(<4 x float> %B, <4 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v4f32.p0f32(<4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st1x2(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st1x2:
+;CHECK: st1.2s { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v2f32.p0f32(<2 x float> %B, <2 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st1x2(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st1x2:
+;CHECK: st1.2s { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v2f32.p0f32(<2 x float> %B, <2 x float> %C, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v2f32.p0f32(<2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st1x2(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st1x2:
+;CHECK: st1.2d { v0, v1 }, [x0], #32
+  call void @llvm.arm64.neon.st1x2.v2f64.p0f64(<2 x double> %B, <2 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st1x2(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st1x2:
+;CHECK: st1.2d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v2f64.p0f64(<2 x double> %B, <2 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v2f64.p0f64(<2 x double>, <2 x double>, double*)
+
+
+define double* @test_v1f64_post_imm_st1x2(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st1x2:
+;CHECK: st1.1d { v0, v1 }, [x0], #16
+  call void @llvm.arm64.neon.st1x2.v1f64.p0f64(<1 x double> %B, <1 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 2
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st1x2(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st1x2:
+;CHECK: st1.1d { v0, v1 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x2.v1f64.p0f64(<1 x double> %B, <1 x double> %C, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x2.v1f64.p0f64(<1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st1x3(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st1x3:
+;CHECK: st1.16b { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i32 48
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st1x3(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st1x3:
+;CHECK: st1.16b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st1x3(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st1x3:
+;CHECK: st1.8b { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i32 24
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st1x3(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st1x3:
+;CHECK: st1.8b { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st1x3(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st1x3:
+;CHECK: st1.8h { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i32 24
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st1x3(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st1x3:
+;CHECK: st1.8h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st1x3(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st1x3:
+;CHECK: st1.4h { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i32 12
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st1x3(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st1x3:
+;CHECK: st1.4h { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, i16*)
+
+
+define i32* @test_v4i32_post_imm_st1x3(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st1x3:
+;CHECK: st1.4s { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i32 12
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st1x3(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st1x3:
+;CHECK: st1.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, i32*)
+
+
+define i32* @test_v2i32_post_imm_st1x3(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st1x3:
+;CHECK: st1.2s { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i32 6
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st1x3(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st1x3:
+;CHECK: st1.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st1x3(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st1x3:
+;CHECK: st1.2d { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 6
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st1x3(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st1x3:
+;CHECK: st1.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, i64*)
+
+
+define i64* @test_v1i64_post_imm_st1x3(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st1x3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 3
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st1x3(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st1x3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, i64*)
+
+
+define float* @test_v4f32_post_imm_st1x3(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st1x3:
+;CHECK: st1.4s { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i32 12
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st1x3(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st1x3:
+;CHECK: st1.4s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st1x3(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st1x3:
+;CHECK: st1.2s { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i32 6
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st1x3(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st1x3:
+;CHECK: st1.2s { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st1x3(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st1x3:
+;CHECK: st1.2d { v0, v1, v2 }, [x0], #48
+  call void @llvm.arm64.neon.st1x3.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 6
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st1x3(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st1x3:
+;CHECK: st1.2d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, double*)
+
+
+define double* @test_v1f64_post_imm_st1x3(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st1x3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], #24
+  call void @llvm.arm64.neon.st1x3.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 3
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st1x3(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st1x3:
+;CHECK: st1.1d { v0, v1, v2 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x3.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x3.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st1x4(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st1x4:
+;CHECK: st1.16b { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i32 64
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st1x4(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st1x4:
+;CHECK: st1.16b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, i8*)
+
+
+define i8* @test_v8i8_post_imm_st1x4(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st1x4:
+;CHECK: st1.8b { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i32 32
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st1x4(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st1x4:
+;CHECK: st1.8b { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>, i8*)
+
+
+define i16* @test_v8i16_post_imm_st1x4(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st1x4:
+;CHECK: st1.8h { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i32 32
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st1x4(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st1x4:
+;CHECK: st1.8h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>, i16*)
+
+
+define i16* @test_v4i16_post_imm_st1x4(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st1x4:
+;CHECK: st1.4h { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i32 16
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st1x4(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st1x4:
+;CHECK: st1.4h { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>,<4 x i16>,  i16*)
+
+
+define i32* @test_v4i32_post_imm_st1x4(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st1x4:
+;CHECK: st1.4s { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i32 16
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st1x4(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st1x4:
+;CHECK: st1.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>,<4 x i32>,  i32*)
+
+
+define i32* @test_v2i32_post_imm_st1x4(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st1x4:
+;CHECK: st1.2s { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i32 8
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st1x4(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st1x4:
+;CHECK: st1.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, i32*)
+
+
+define i64* @test_v2i64_post_imm_st1x4(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st1x4:
+;CHECK: st1.2d { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 8
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st1x4(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st1x4:
+;CHECK: st1.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>,<2 x i64>,  i64*)
+
+
+define i64* @test_v1i64_post_imm_st1x4(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st1x4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st1x4(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st1x4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>,<1 x i64>,  i64*)
+
+
+define float* @test_v4f32_post_imm_st1x4(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st1x4:
+;CHECK: st1.4s { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i32 16
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st1x4(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st1x4:
+;CHECK: st1.4s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, <4 x float>, float*)
+
+
+define float* @test_v2f32_post_imm_st1x4(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st1x4:
+;CHECK: st1.2s { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i32 8
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st1x4(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st1x4:
+;CHECK: st1.2s { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, <2 x float>, float*)
+
+
+define double* @test_v2f64_post_imm_st1x4(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st1x4:
+;CHECK: st1.2d { v0, v1, v2, v3 }, [x0], #64
+  call void @llvm.arm64.neon.st1x4.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 8
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st1x4(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st1x4:
+;CHECK: st1.2d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>,<2 x double>,  double*)
+
+
+define double* @test_v1f64_post_imm_st1x4(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st1x4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], #32
+  call void @llvm.arm64.neon.st1x4.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st1x4(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st1x4:
+;CHECK: st1.1d { v0, v1, v2, v3 }, [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st1x4.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st1x4.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, <1 x double>, double*)
+
+
+define i8* @test_v16i8_post_imm_st2lanelane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C) {
+  call void @llvm.arm64.neon.st2lanelane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i64 1, i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st2lanelane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, i64 %inc) {
+  call void @llvm.arm64.neon.st2lanelane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i64 1, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lanelane.v16i8.p0i8(<16 x i8>, <16 x i8>, i64, i64, i8*) nounwind readnone
+
+
+define i8* @test_v16i8_post_imm_st2lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st2lane:
+;CHECK: st2.b { v0, v1 }[0], [x0], #2
+  call void @llvm.arm64.neon.st2lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st2lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st2lane:
+;CHECK: st2.b { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v16i8.p0i8(<16 x i8>, <16 x i8>, i64, i8*)
+
+
+define i8* @test_v8i8_post_imm_st2lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st2lane:
+;CHECK: st2.b { v0, v1 }[0], [x0], #2
+  call void @llvm.arm64.neon.st2lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 2
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st2lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st2lane:
+;CHECK: st2.b { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v8i8.p0i8(<8 x i8>, <8 x i8>, i64, i8*)
+
+
+define i16* @test_v8i16_post_imm_st2lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st2lane:
+;CHECK: st2.h { v0, v1 }[0], [x0], #4
+  call void @llvm.arm64.neon.st2lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st2lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st2lane:
+;CHECK: st2.h { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v8i16.p0i16(<8 x i16>, <8 x i16>, i64, i16*)
+
+
+define i16* @test_v4i16_post_imm_st2lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st2lane:
+;CHECK: st2.h { v0, v1 }[0], [x0], #4
+  call void @llvm.arm64.neon.st2lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 2
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st2lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st2lane:
+;CHECK: st2.h { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v4i16.p0i16(<4 x i16>, <4 x i16>, i64, i16*)
+
+
+define i32* @test_v4i32_post_imm_st2lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], #8
+  call void @llvm.arm64.neon.st2lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st2lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v4i32.p0i32(<4 x i32>, <4 x i32>, i64, i32*)
+
+
+define i32* @test_v2i32_post_imm_st2lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], #8
+  call void @llvm.arm64.neon.st2lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 2
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st2lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v2i32.p0i32(<2 x i32>, <2 x i32>, i64, i32*)
+
+
+define i64* @test_v2i64_post_imm_st2lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], #16
+  call void @llvm.arm64.neon.st2lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 2
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st2lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v2i64.p0i64(<2 x i64>, <2 x i64>, i64, i64*)
+
+
+define i64* @test_v1i64_post_imm_st2lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], #16
+  call void @llvm.arm64.neon.st2lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 2
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st2lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v1i64.p0i64(<1 x i64>, <1 x i64>, i64, i64*)
+
+
+define float* @test_v4f32_post_imm_st2lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], #8
+  call void @llvm.arm64.neon.st2lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st2lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v4f32.p0f32(<4 x float>, <4 x float>, i64, float*)
+
+
+define float* @test_v2f32_post_imm_st2lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], #8
+  call void @llvm.arm64.neon.st2lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 2
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st2lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st2lane:
+;CHECK: st2.s { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v2f32.p0f32(<2 x float>, <2 x float>, i64, float*)
+
+
+define double* @test_v2f64_post_imm_st2lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], #16
+  call void @llvm.arm64.neon.st2lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 2
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st2lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v2f64.p0f64(<2 x double>, <2 x double>, i64, double*)
+
+
+define double* @test_v1f64_post_imm_st2lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], #16
+  call void @llvm.arm64.neon.st2lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 2
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st2lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st2lane:
+;CHECK: st2.d { v0, v1 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st2lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st2lane.v1f64.p0f64(<1 x double>, <1 x double>, i64, double*)
+
+
+define i8* @test_v16i8_post_imm_st3lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st3lane:
+;CHECK: st3.b { v0, v1, v2 }[0], [x0], #3
+  call void @llvm.arm64.neon.st3lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st3lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st3lane:
+;CHECK: st3.b { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, i64, i8*)
+
+
+define i8* @test_v8i8_post_imm_st3lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st3lane:
+;CHECK: st3.b { v0, v1, v2 }[0], [x0], #3
+  call void @llvm.arm64.neon.st3lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 3
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st3lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st3lane:
+;CHECK: st3.b { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, i64, i8*)
+
+
+define i16* @test_v8i16_post_imm_st3lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st3lane:
+;CHECK: st3.h { v0, v1, v2 }[0], [x0], #6
+  call void @llvm.arm64.neon.st3lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st3lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st3lane:
+;CHECK: st3.h { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, i64, i16*)
+
+
+define i16* @test_v4i16_post_imm_st3lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st3lane:
+;CHECK: st3.h { v0, v1, v2 }[0], [x0], #6
+  call void @llvm.arm64.neon.st3lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 3
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st3lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st3lane:
+;CHECK: st3.h { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, i64, i16*)
+
+
+define i32* @test_v4i32_post_imm_st3lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], #12
+  call void @llvm.arm64.neon.st3lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st3lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, i64, i32*)
+
+
+define i32* @test_v2i32_post_imm_st3lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], #12
+  call void @llvm.arm64.neon.st3lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 3
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st3lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, i64, i32*)
+
+
+define i64* @test_v2i64_post_imm_st3lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], #24
+  call void @llvm.arm64.neon.st3lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 3
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st3lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, i64, i64*)
+
+
+define i64* @test_v1i64_post_imm_st3lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], #24
+  call void @llvm.arm64.neon.st3lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 3
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st3lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, i64, i64*)
+
+
+define float* @test_v4f32_post_imm_st3lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], #12
+  call void @llvm.arm64.neon.st3lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st3lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, i64, float*)
+
+
+define float* @test_v2f32_post_imm_st3lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], #12
+  call void @llvm.arm64.neon.st3lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 3
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st3lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st3lane:
+;CHECK: st3.s { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, i64, float*)
+
+
+define double* @test_v2f64_post_imm_st3lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], #24
+  call void @llvm.arm64.neon.st3lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 3
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st3lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, i64, double*)
+
+
+define double* @test_v1f64_post_imm_st3lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], #24
+  call void @llvm.arm64.neon.st3lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 3
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st3lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st3lane:
+;CHECK: st3.d { v0, v1, v2 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st3lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st3lane.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, i64, double*)
+
+
+define i8* @test_v16i8_post_imm_st4lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E) nounwind {
+;CHECK-LABEL: test_v16i8_post_imm_st4lane:
+;CHECK: st4.b { v0, v1, v2, v3 }[0], [x0], #4
+  call void @llvm.arm64.neon.st4lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  ret i8* %tmp
+}
+
+define i8* @test_v16i8_post_reg_st4lane(i8* %A, i8** %ptr, <16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v16i8_post_reg_st4lane:
+;CHECK: st4.b { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v16i8.p0i8(<16 x i8> %B, <16 x i8> %C, <16 x i8> %D, <16 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v16i8.p0i8(<16 x i8>, <16 x i8>, <16 x i8>, <16 x i8>, i64, i8*)
+
+
+define i8* @test_v8i8_post_imm_st4lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E) nounwind {
+;CHECK-LABEL: test_v8i8_post_imm_st4lane:
+;CHECK: st4.b { v0, v1, v2, v3 }[0], [x0], #4
+  call void @llvm.arm64.neon.st4lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i32 4
+  ret i8* %tmp
+}
+
+define i8* @test_v8i8_post_reg_st4lane(i8* %A, i8** %ptr, <8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i8_post_reg_st4lane:
+;CHECK: st4.b { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v8i8.p0i8(<8 x i8> %B, <8 x i8> %C, <8 x i8> %D, <8 x i8> %E, i64 0, i8* %A)
+  %tmp = getelementptr i8* %A, i64 %inc
+  ret i8* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v8i8.p0i8(<8 x i8>, <8 x i8>, <8 x i8>, <8 x i8>, i64, i8*)
+
+
+define i16* @test_v8i16_post_imm_st4lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E) nounwind {
+;CHECK-LABEL: test_v8i16_post_imm_st4lane:
+;CHECK: st4.h { v0, v1, v2, v3 }[0], [x0], #8
+  call void @llvm.arm64.neon.st4lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  ret i16* %tmp
+}
+
+define i16* @test_v8i16_post_reg_st4lane(i16* %A, i16** %ptr, <8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v8i16_post_reg_st4lane:
+;CHECK: st4.h { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v8i16.p0i16(<8 x i16> %B, <8 x i16> %C, <8 x i16> %D, <8 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v8i16.p0i16(<8 x i16>, <8 x i16>, <8 x i16>, <8 x i16>, i64, i16*)
+
+
+define i16* @test_v4i16_post_imm_st4lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E) nounwind {
+;CHECK-LABEL: test_v4i16_post_imm_st4lane:
+;CHECK: st4.h { v0, v1, v2, v3 }[0], [x0], #8
+  call void @llvm.arm64.neon.st4lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i32 4
+  ret i16* %tmp
+}
+
+define i16* @test_v4i16_post_reg_st4lane(i16* %A, i16** %ptr, <4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i16_post_reg_st4lane:
+;CHECK: st4.h { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v4i16.p0i16(<4 x i16> %B, <4 x i16> %C, <4 x i16> %D, <4 x i16> %E, i64 0, i16* %A)
+  %tmp = getelementptr i16* %A, i64 %inc
+  ret i16* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v4i16.p0i16(<4 x i16>, <4 x i16>, <4 x i16>, <4 x i16>, i64, i16*)
+
+
+define i32* @test_v4i32_post_imm_st4lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E) nounwind {
+;CHECK-LABEL: test_v4i32_post_imm_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], #16
+  call void @llvm.arm64.neon.st4lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  ret i32* %tmp
+}
+
+define i32* @test_v4i32_post_reg_st4lane(i32* %A, i32** %ptr, <4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4i32_post_reg_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v4i32.p0i32(<4 x i32> %B, <4 x i32> %C, <4 x i32> %D, <4 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v4i32.p0i32(<4 x i32>, <4 x i32>, <4 x i32>, <4 x i32>, i64, i32*)
+
+
+define i32* @test_v2i32_post_imm_st4lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E) nounwind {
+;CHECK-LABEL: test_v2i32_post_imm_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], #16
+  call void @llvm.arm64.neon.st4lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i32 4
+  ret i32* %tmp
+}
+
+define i32* @test_v2i32_post_reg_st4lane(i32* %A, i32** %ptr, <2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i32_post_reg_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v2i32.p0i32(<2 x i32> %B, <2 x i32> %C, <2 x i32> %D, <2 x i32> %E, i64 0, i32* %A)
+  %tmp = getelementptr i32* %A, i64 %inc
+  ret i32* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v2i32.p0i32(<2 x i32>, <2 x i32>, <2 x i32>, <2 x i32>, i64, i32*)
+
+
+define i64* @test_v2i64_post_imm_st4lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E) nounwind {
+;CHECK-LABEL: test_v2i64_post_imm_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], #32
+  call void @llvm.arm64.neon.st4lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v2i64_post_reg_st4lane(i64* %A, i64** %ptr, <2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2i64_post_reg_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v2i64.p0i64(<2 x i64> %B, <2 x i64> %C, <2 x i64> %D, <2 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v2i64.p0i64(<2 x i64>, <2 x i64>, <2 x i64>, <2 x i64>, i64, i64*)
+
+
+define i64* @test_v1i64_post_imm_st4lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E) nounwind {
+;CHECK-LABEL: test_v1i64_post_imm_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], #32
+  call void @llvm.arm64.neon.st4lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 4
+  ret i64* %tmp
+}
+
+define i64* @test_v1i64_post_reg_st4lane(i64* %A, i64** %ptr, <1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1i64_post_reg_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v1i64.p0i64(<1 x i64> %B, <1 x i64> %C, <1 x i64> %D, <1 x i64> %E, i64 0, i64* %A)
+  %tmp = getelementptr i64* %A, i64 %inc
+  ret i64* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v1i64.p0i64(<1 x i64>, <1 x i64>, <1 x i64>, <1 x i64>, i64, i64*)
+
+
+define float* @test_v4f32_post_imm_st4lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E) nounwind {
+;CHECK-LABEL: test_v4f32_post_imm_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], #16
+  call void @llvm.arm64.neon.st4lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  ret float* %tmp
+}
+
+define float* @test_v4f32_post_reg_st4lane(float* %A, float** %ptr, <4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v4f32_post_reg_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v4f32.p0f32(<4 x float> %B, <4 x float> %C, <4 x float> %D, <4 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v4f32.p0f32(<4 x float>, <4 x float>, <4 x float>, <4 x float>, i64, float*)
+
+
+define float* @test_v2f32_post_imm_st4lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E) nounwind {
+;CHECK-LABEL: test_v2f32_post_imm_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], #16
+  call void @llvm.arm64.neon.st4lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i32 4
+  ret float* %tmp
+}
+
+define float* @test_v2f32_post_reg_st4lane(float* %A, float** %ptr, <2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f32_post_reg_st4lane:
+;CHECK: st4.s { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v2f32.p0f32(<2 x float> %B, <2 x float> %C, <2 x float> %D, <2 x float> %E, i64 0, float* %A)
+  %tmp = getelementptr float* %A, i64 %inc
+  ret float* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v2f32.p0f32(<2 x float>, <2 x float>, <2 x float>, <2 x float>, i64, float*)
+
+
+define double* @test_v2f64_post_imm_st4lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E) nounwind {
+;CHECK-LABEL: test_v2f64_post_imm_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], #32
+  call void @llvm.arm64.neon.st4lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v2f64_post_reg_st4lane(double* %A, double** %ptr, <2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v2f64_post_reg_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v2f64.p0f64(<2 x double> %B, <2 x double> %C, <2 x double> %D, <2 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v2f64.p0f64(<2 x double>, <2 x double>, <2 x double>, <2 x double>, i64, double*)
+
+
+define double* @test_v1f64_post_imm_st4lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E) nounwind {
+;CHECK-LABEL: test_v1f64_post_imm_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], #32
+  call void @llvm.arm64.neon.st4lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 4
+  ret double* %tmp
+}
+
+define double* @test_v1f64_post_reg_st4lane(double* %A, double** %ptr, <1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 %inc) nounwind {
+;CHECK-LABEL: test_v1f64_post_reg_st4lane:
+;CHECK: st4.d { v0, v1, v2, v3 }[0], [x0], x{{[0-9]+}}
+  call void @llvm.arm64.neon.st4lane.v1f64.p0f64(<1 x double> %B, <1 x double> %C, <1 x double> %D, <1 x double> %E, i64 0, double* %A)
+  %tmp = getelementptr double* %A, i64 %inc
+  ret double* %tmp
+}
+
+declare void @llvm.arm64.neon.st4lane.v1f64.p0f64(<1 x double>, <1 x double>, <1 x double>, <1 x double>, i64, double*)
\ No newline at end of file